%0 Conference Paper %1 ODH20 %A Osband, Ian %A Doron, Yotam %A Hessel, Matteo %A Aslanides, John %A Sezener, Eren %A Saraiva, Andre %A McKinney, Katrina %A Lattimore, Tor %A Szepesvári, Csaba %A Singh, Satinder %A Roy, Benjamin Van %A Sutton, Richard S. %A Silver, David %A van Hasselt, Hado %B ICLR %D 2020 %K empirical evaluation learning, reinforcement %T Behaviour Suite for Reinforcement Learning %U http://arxiv.org/abs/1908.03568 %X This paper introduces the Behaviour Suite for Reinforcement Learning, or bsuite for short. bsuite is a collection of carefully-designed experiments that investigate core capabilities of reinforcement learning (RL) agents with two objectives. First, to collect clear, informative and scalable problems that capture key issues in the design of general and efficient learning algorithms. Second, to study agent behaviour through their performance on these shared benchmarks. To complement this effort, we open source this, which automates evaluation and analysis of any agent on bsuite. This library facilitates reproducible and accessible research on the core issues in RL, and ultimately the design of superior learning algorithms. Our code is Python, and easy to use within existing projects. We include examples with OpenAI Baselines, Dopamine as well as new reference implementations. Going forward, we hope to incorporate more excellent experiments from the research community, and commit to a periodic review of bsuite from a committee of prominent researchers.

@inproceedings{ODH20, abstract = {This paper introduces the Behaviour Suite for Reinforcement Learning, or bsuite for short. bsuite is a collection of carefully-designed experiments that investigate core capabilities of reinforcement learning (RL) agents with two objectives. First, to collect clear, informative and scalable problems that capture key issues in the design of general and efficient learning algorithms. Second, to study agent behaviour through their performance on these shared benchmarks. To complement this effort, we open source this, which automates evaluation and analysis of any agent on bsuite. This library facilitates reproducible and accessible research on the core issues in RL, and ultimately the design of superior learning algorithms. Our code is Python, and easy to use within existing projects. We include examples with OpenAI Baselines, Dopamine as well as new reference implementations. Going forward, we hope to incorporate more excellent experiments from the research community, and commit to a periodic review of bsuite from a committee of prominent researchers.}, added-at = {2020-03-17T03:03:01.000+0100}, author = {Osband, Ian and Doron, Yotam and Hessel, Matteo and Aslanides, John and Sezener, Eren and Saraiva, Andre and McKinney, Katrina and Lattimore, Tor and Szepesv\'{a}ri, Csaba and Singh, Satinder and Roy, Benjamin Van and Sutton, Richard S. and Silver, David and van Hasselt, Hado}, bdsk-url-1 = {http://arxiv.org/abs/1908.03568}, biburl = {https://www.bibsonomy.org/bibtex/2f079298f6bcc2d70c8d603ace0bf55c6/csaba}, booktitle = {ICLR}, date-added = {2020-03-08 16:04:53 -0600}, date-modified = {2020-03-08 16:07:01 -0600}, interhash = {231e144bf529d835d683a77d576aa667}, intrahash = {f079298f6bcc2d70c8d603ace0bf55c6}, keywords = {empirical evaluation learning, reinforcement}, timestamp = {2020-03-17T03:03:01.000+0100}, title = {Behaviour Suite for Reinforcement Learning}, url = {http://arxiv.org/abs/1908.03568}, year = 2020 }